Intelligent control module implementation method based on distributed control system

Abstract

The invention relates to an intelligent control module implementation method based on a decentralized control system, which comprises the following steps of: developing an intelligent control algorithm program, and establishing a matrix variable to represent a model parameter required by an algorithm; the program input comprises model parameter update variables and control parameters; the optimal model parameters calculated off-line through the optimization algorithm are stored as data files and stored in a set path; an intelligent control algorithm program is loaded in the decentralized control system, corresponding data files are read respectively, and model parameters are set; one intelligent control algorithm program is selected to be output as a final control instruction of the multiple intelligent control algorithm programs, and the running state of each intelligent control algorithm program is judged according to the value of the model parameter updating variable; when the program is in a normal operation state, obtaining input to output a control instruction; when the program is in the model loading state, the latest data file in the corresponding set path is read again, the current model parameter is updated, and in the updating process, the control instruction output by the module in the normal operation state is tracked. Compared with the prior art, the method has the advantages that intelligent control algorithm model parameters can be updated online, and the output control instruction is more effective.

Description

Technical Field

[0001] The present invention relates to the technical field of power generation automatic control, and in particular to a method for realizing an intelligent control module based on a distributed control system. Background Art

[0002] The traditional distributed control system (DCS) is limited by the closed system and insufficient computing power. It can only implement conventional control methods and cannot effectively solve many control problems of thermal power units. In order to further improve the control performance of thermal power units, intelligent control has been gradually applied to the control field of thermal power units. The application of intelligent control methods in traditional DCS can only be done by PLC or server "plug-in", which has many problems such as poor openness, difficult operation and maintenance, inconsistent standards, and chaotic systems.

[0003] At present, with the product update and promotion of DCS, the research and application of intelligent control algorithms have gradually begun based on the characteristics of independent development, flexible application and high-performance controllers. However, due to factors such as equipment aging and system transformation, the characteristics of generator set equipment and systems will gradually change, and intelligent control algorithms are usually implemented based on more accurate model parameters. In order to prevent the control performance of the control module from deteriorating due to model parameter mismatch, the model parameters of the algorithm need to be updated from time to time. This update operation requires recompiling and downloading the controller. For the sake of safe operation, it is usually necessary to perform it during the unit shutdown period. If the model parameters are not updated, the control instructions are calculated using mismatched model parameters, which will affect the stability and economy of the unit's automatic control system. At the same time, because DCS needs to meet real-time requirements and limited computing power, complex algorithms such as model identification and neural network training are difficult to carry out in real time in DCS, and the actual application effect of intelligent control algorithms is not good. Summary of the invention

[0004] The purpose of the present invention is to provide a method for realizing an intelligent control module based on a distributed control system. In view of the problems that the model parameters of the intelligent control module embedded in the DCS are difficult to update, resulting in poor practical application, the present invention adopts a model parameter file transmission method, generates corresponding files according to the model parameters of each intelligent control algorithm program, saves them in an independent loading area, and cooperates with the model parameter update variables to flexibly load the model parameters online, thereby realizing the online update of the intelligent control model parameters, ensuring the efficient application of the intelligent control algorithm, and improving the intelligent control level of the unit.

[0005] The purpose of the present invention can be achieved by the following technical solutions:

[0006] A method for implementing an intelligent control module based on a distributed control system comprises the following steps:

[0007] Develop an intelligent control algorithm program, and use matrix variables to represent the model parameters required by the intelligent control algorithm. The intelligent control algorithm program takes the control parameters and the model parameter update variables as inputs based on the intelligent control model parameters, and calculates the output control instructions;

[0008] The optimal intelligent control model parameters calculated offline by the optimization algorithm are saved as data files and stored in the set path;

[0009] In a distributed control system, multiple intelligent control algorithm programs load corresponding intelligent control model parameters by reading matrix variables in data files under set paths respectively;

[0010] Each intelligent control algorithm program determines the program running state according to the value of the variable updated by the respective model parameter, and the program running state includes a normal running state and a model loading state;

[0011] When the program is in normal operation, control instructions are output according to the obtained model parameters and input control parameters;

[0012] When the program is in the state of loading the model, it re-reads the latest data file in the set path, updates the model parameters according to the matrix variables in the data file, and tracks the control instructions output by other intelligent control algorithm programs in normal operation during the process of updating the model parameters;

[0013] One of the control instructions output by multiple intelligent control algorithm programs is selected as the final control output.

[0014] The model parameter update variable is of Boolean type. When its value is true, the intelligent control model parameters need to be updated and the program enters the model loading state. When its value is false, the intelligent control model parameters do not need to be updated and the program is in normal operation.

[0015] The data file is updated in an overwrite mode. When a data file is updated, the original data file is overwritten in the corresponding path.

[0016] When the program enters the model loading state for a preset time, the model parameters are updated and the values ​​of the variables are updated so that the program enters the normal operation state, and the control instructions are output according to the updated intelligent control model parameters.

[0017] In the development environment of the distributed control system, the intelligent control algorithm program is compiled into a dynamic link library file and packaged into a graphical intelligent control module according to the packaging rules. The input of the program is packaged as the input pin in the corresponding graphical intelligent control module. The module performs calculations by loading the dynamic link library file, reads the input required by the program by obtaining the value of the input pin, and outputs it through the output pin.

[0018] The graphical intelligent control module reads the module ID number assigned to itself by the distributed control system, and automatically establishes a corresponding folder in the controller as the set path based on the module ID number; when multiple graphical intelligent control modules are added to the distributed control system, the modules read the data files under their respective paths through the set path generated by the module ID number.

[0019] When the method is applied to the generator set outlet nitrogen oxide PID control based on the particle swarm optimization algorithm, the model parameters include matrix variables composed of the proportional parameter, integral parameter and differential parameter of each particle, and the control parameters include the nitrogen oxide concentration measurement value, the nitrogen oxide concentration set value, the tracking value, the manual automatic state, the cognitive factor, the social factor, the number of particles, and the upper and lower limits of the proportional parameter, integral parameter and differential parameter in the particle parameters.

[0020] The manual state tracking value is the actual opening of the ammonia injection valve to be tracked.

[0021] The model parameter update variables are set by manual adjustment.

[0022] The intelligent algorithms include genetic algorithms, reinforcement learning algorithms, regression algorithms, ensemble learning algorithms, deep learning algorithms and particle swarm optimization algorithms.

[0023] Compared with the prior art, the present invention has the following beneficial effects:

[0024] In view of the problem that it is difficult to update model parameters online when a DCS embedded intelligent control algorithm is used, the present invention adopts a model parameter file transmission method to save the intelligent control model parameters used by each graphical intelligent control module in the form of a file to an independent loading area, and in combination with the setting of model parameter update variables, the model parameters can be flexibly loaded, thereby realizing the online update of the intelligent control model parameters; and during the model parameter update, other intelligent control modules that are not being updated are used to replace the output, so that the output control instructions can temporarily maintain real-time tracking, reduce safety hazards, and provide strong support for improving the automation level of the unit. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 is a flow chart of the method of the present invention;

[0026] Figure 2 A schematic diagram of input and output pins of a graphical intelligent control module in an embodiment. DETAILED DESCRIPTION

[0027] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. This embodiment is implemented based on the technical solution of the present invention, and provides a detailed implementation method and specific operation process, but the protection scope of the present invention is not limited to the following embodiments.

[0028] Example 1

[0029] This embodiment provides a method for implementing an intelligent control module based on a distributed control system. Figure 1 As shown, the following steps are included:

[0030] S1, develop an intelligent control algorithm program, and use matrix variables to represent the model parameters required by the intelligent control algorithm. The intelligent control algorithm program calculates and outputs control instructions based on the intelligent control model parameters, control parameters and model parameter update variables.

[0031] This embodiment develops an intelligent control algorithm program according to the development specification of the distributed control system, and establishes a matrix variable G as a representation of the intelligent control model parameters. The program loads G by reading a data file under a set path.

[0032] Add a special input signal in the intelligent control algorithm program - model parameter update variable s. Model parameter update variable s is a Boolean type, with a value of True or False. When its value is True, the intelligent control model parameters need to be updated and enter the loading model state; when its value is False, the intelligent control model parameters do not need to be updated and are in normal operation.

[0033] In this embodiment, the model parameter update variable is manually set according to actual needs. In other embodiments, an automatic setting method can also be adopted. For example, when it is detected that the data file update is completed, the value of the model parameter update variable is automatically changed from False to True.

[0034] In this embodiment, the intelligent control algorithm includes but is not limited to a genetic algorithm, a reinforcement learning algorithm, a regression algorithm, an ensemble learning algorithm, a deep learning algorithm, and a particle swarm optimization algorithm, and such intelligent control algorithms include a large number of parameters that need to be trained. The corresponding model parameters represent the necessary parameters in the adopted intelligent control algorithm, and the control parameters represent the physical quantity parameters related to the control object and application scenario of the intelligent control.

[0035] S2, saves the best model parameters calculated offline by the optimization algorithm as a data file in the form of matrix variables and stores it in the set path.

[0036] In this embodiment, the data file is updated in an overwrite mode. When the data file is updated, the original data file is overwritten in the set path. After the new intelligent control model parameters are calculated offline by the optimization algorithm, the model parameters are saved as a data file and updated to the data file in the set path.

[0037] S3, in the distributed control system, multiple intelligent control algorithm programs load corresponding intelligent control model parameters by reading matrix variables in the data file under the set path respectively. In the distributed control system development environment, the intelligent control algorithm program is compiled into a dynamic link library file, and packaged into a graphical intelligent control module according to the packaging rules. The input of the program is packaged as the input pin in the corresponding graphical intelligent control module. The module performs calculations by loading the dynamic link library file, reads the input required by the program by obtaining the value of the input pin, and outputs through the output pin.

[0038] The graphical intelligent control module reads the module ID number assigned to it by the distributed control system, and automatically creates a corresponding folder in the controller as the set path according to the module ID number. When multiple graphical intelligent control modules are added to the distributed control system, the set path generated by the module ID number reads the data files under each path to avoid confusion in the reading file path and ensure that the data file is uniquely corresponding to the algorithm module.

[0039] S4, adding multiple graphical intelligent control modules in the distributed control system, each graphical intelligent control module updates the value of the variable according to the model parameter input by itself to determine its own program running status, and the program running status includes normal running status (s=False) and loading model status (s=True).

[0040] S5, when the program is in normal operation, output control instructions according to the obtained model parameters and input control parameters.

[0041] S6, when the program is in the state of loading the model, re-read the latest data file under the set path, update the model parameters according to the matrix variables in the data file, and track the control instructions output by other intelligent control algorithm programs in normal operation during the process of updating the model parameters.

[0042] In this embodiment, when the data file under the set path is updated, the input pin corresponding to the model parameter update variable s of the intelligent control module is set to True, so that the program enters the model loading state.

[0043] When the program enters the loading model state for 10 seconds, it is considered that the model parameter update has been completed, and the value of the model parameter update variable is updated to False, so that the program enters the normal operation state and outputs control instructions according to the updated intelligent control model parameters.

[0044] After completing the online update of the intelligent control model parameters, the module enters the normal control process. During the online operation process, if it is found that the control instructions output by the module are difficult to meet the control requirements, the value of the model parameter update variable s can be set to make it enter step S6 again for online update without compiling and downloading the controller.

[0045] S7, selecting a control instruction output by a graphical intelligent control module as the final control output.

[0046] In this embodiment, one of the multiple intelligent control modules is selected as the final output by manual setting. In other embodiments, an automatic judgment method can also be adopted. For example, an automatic judgment program is added to evaluate the output of each intelligent control module in real time and select the best one as the final output.

[0047] Example 2

[0048] This embodiment takes the generator set outlet nitrogen oxide PID control based on the particle swarm optimization algorithm as an example to further illustrate the method of Embodiment 1.

[0049] The NOx control system at the outlet of the denitrification system of a 660MW generator set originally adopted a conventional PID control strategy. Affected by factors such as NOx measurement lag and frequent and drastic changes in unit operating conditions, NOx control fluctuated greatly and often exceeded the standard. In order to improve the NOx control effect, a particle swarm intelligent PID algorithm is implemented based on a distributed control system. Since the characteristics of the boiler air and smoke system and the denitrification catalytic reaction system will change with the increase of operating time, the model parameters of the particle swarm intelligent PID algorithm do not match the actual situation after running for a period of time. The model parameters need to be updated during the unit shutdown period. The model parameters are not updated in time, resulting in poor actual operation results. The particle swarm intelligent PID algorithm is implemented using the technical solution of the present invention, and the implementation process is as follows:

[0050] (1) According to the distributed control system development specification, the particle swarm intelligent PID algorithm program of the denitrification control system is developed, and the proportion parameters P1,…,P of the m particles required by the particle swarm algorithm are m , integration parameters I1,…,I m and differential parameters D1,…,D m Establish a parameter matrix variable G, and G is loaded by reading the data file under the set path.

[0051] (2) Add parameter update variable s as a special input signal in the particle swarm intelligent PID algorithm program and perform logical judgment on s. The program needs to read the input parameters required by the particle swarm intelligent PID algorithm, such as the measured value PVin, the set value SPin, the tracking value TR, the cognitive factor C1, the social factor C2, the number of particles m, and the particle parameters P1 to P m Upper limit PH, particle parameters P1 to P m Lower limit PL, particle parameters I1 to I m Upper limit IH, particle parameters I1 to I m Lower limit IL, particle parameters D1 to Dm Upper limit DH, particle parameters D1 to D m The lower limit DL, the manual and automatic state STR, and the value of the model parameter update variable s, s takes the value of True or False. When the value of s is False, the program switches to the normal operation state, performs operations based on the parameter matrix variable G, and outputs the operation results as control instructions; when the value of s is True, the program switches to the loading model state, tracks the control instructions output by other intelligent control algorithm programs in normal operation, and at the same time, the program reads the data file under the set path, assigns the read data to the matrix variable G, and updates the large-scale model parameters in the algorithm online.

[0052] (3) In the distributed control system development environment, the particle swarm intelligent PID algorithm program is compiled into a dynamic link library file and encapsulated into a graphical module according to the encapsulation rules. The input and output of the program are encapsulated as the input and output pins in the corresponding graphical module. The module performs calculations by loading the dynamic link library file and reads the input required by the program by obtaining the value of the input pin.

[0053] (4) Add the particle swarm intelligent PID module in the distributed control system configuration interface and connect the input and output pins. PVin is connected to the actual value of the outlet Nox concentration, SPin is connected to the set value of the outlet Nox concentration, TR is connected to the output of other intelligent control modules to be tracked, C1 and C2 are set to 2, m is set to 100, PH is set to 2, PL is set to 0.1, IH is set to 0.5, IL is set to 0.01, DH is set to 0.1, DL is set to 0, and STR is connected to the manual-automatic switching command. Figure 2 As shown, the module ID number assigned by DCS is 1. The module creates a folder named "DATA1" in the controller as the setting path based on the ID number.

[0054] (5) After obtaining the parameter matrix variable G through offline calculation, save G as a data file and update it to the "DATA1" folder. Set the input pin corresponding to the module variable s to True and set it to False after 10 seconds. Update the model parameters required by the particle swarm intelligent PID algorithm online.

[0055] (6) After the online update is completed, the particle swarm intelligent PID module enters the normal control process. During the online commissioning process, step (5) can be repeated to perform online update again without compiling and downloading the controller.

[0056] (7) Select the output of one of the multiple particle swarm intelligent PID modules as the final control output.

[0057] The preferred specific embodiments of the present invention are described in detail above. It should be understood that a person skilled in the art can make many modifications and changes based on the concept of the present invention without creative work. Therefore, any technical solution that can be obtained by a person skilled in the art through logical analysis, reasoning, or limited experiments based on the concept of the present invention on the basis of the prior art should be within the scope of protection determined by the claims.

Claims

1. A method for implementing an intelligent control module based on a distributed control system, characterized in that: The following steps are involved: Develop an intelligent control algorithm program, and use matrix variables to represent the model parameters required by the intelligent control algorithm. The intelligent control algorithm program takes the control parameters and the model parameter update variables as inputs based on the intelligent control model parameters, and calculates the output control instructions; The optimal intelligent control model parameters calculated offline by the optimization algorithm are saved as data files and stored in the set path; In a distributed control system, multiple intelligent control algorithm programs load corresponding intelligent control model parameters by reading matrix variables in data files under set paths respectively; Each intelligent control algorithm program determines the program running state according to the value of the variable updated by the respective model parameter, and the program running state includes a normal running state and a model loading state; When the program is in normal operation, control instructions are output according to the obtained model parameters and input control parameters; When the program is in the state of loading the model, it re-reads the latest data file in the set path, updates the model parameters according to the matrix variables in the data file, and tracks the control instructions output by other intelligent control algorithm programs in normal operation during the process of updating the model parameters; One of the control instructions output by multiple intelligent control algorithm programs is selected as the final control output.

2. The method for realizing an intelligent control module based on a distributed control system according to claim 1, characterized in that: The model parameter update variable is of Boolean type. When its value is true, the intelligent control model parameters need to be updated and the program enters the model loading state. When its value is false, the intelligent control model parameters do not need to be updated and the program is in normal operation.

3. The method for realizing an intelligent control module based on a distributed control system according to claim 1, characterized in that: The data file is updated in an overwrite mode. When a data file is updated, the original data file is overwritten in the corresponding path.

4. The method for realizing an intelligent control module based on a distributed control system according to claim 1, characterized in that: When the program enters the model loading state for a preset time, the model parameters are updated and the values ​​of the variables are updated so that the program enters the normal operation state, and the control instructions are output according to the updated intelligent control model parameters.

5. The method for realizing an intelligent control module based on a distributed control system according to claim 1, characterized in that: In the development environment of the distributed control system, the intelligent control algorithm program is compiled into a dynamic link library file and packaged into a graphical intelligent control module according to the packaging rules. The input of the program is packaged as the input pin in the corresponding graphical intelligent control module. The module performs calculations by loading the dynamic link library file, reads the input required by the program by obtaining the value of the input pin, and outputs it through the output pin.

6. The method for realizing an intelligent control module based on a distributed control system according to claim 5, characterized in that: The graphical intelligent control module reads the module ID number assigned to itself by the distributed control system, and automatically establishes a corresponding folder in the controller as the set path based on the module ID number; when multiple graphical intelligent control modules are added to the distributed control system, the modules read the data files under their respective paths through the set path generated by the module ID number.

7. The method for realizing an intelligent control module based on a distributed control system according to claim 1, characterized in that: When the method is applied to the generator set outlet nitrogen oxide PID control based on the particle swarm optimization algorithm, the model parameters include matrix variables composed of the proportional parameter, integral parameter and differential parameter of each particle, and the control parameters include the nitrogen oxide concentration measurement value, the nitrogen oxide concentration set value, the tracking value, the manual automatic state, the cognitive factor, the social factor, the number of particles, and the upper and lower limits of the proportional parameter, integral parameter and differential parameter in the particle parameters.

8. The method for realizing an intelligent control module based on a distributed control system according to claim 7, characterized in that: The manual state tracking value is the actual opening of the ammonia injection valve to be tracked.

9. The method for realizing an intelligent control module based on a distributed control system according to claim 1, characterized in that: The model parameter update variables are set by manual adjustment.

10. The method for realizing an intelligent control module based on a distributed control system according to claim 1, characterized in that: The intelligent control algorithms include genetic algorithms, reinforcement learning algorithms, regression algorithms, ensemble learning algorithms, deep learning algorithms and particle swarm optimization algorithms.

Images